URL pattern detect

NaiboWang · Jul 5, 2023 · f84eb67 · f84eb67
1 parent a0dbaea
commit f84eb67
Show file tree

Hide file tree

Showing 14 changed files with 150 additions and 60 deletions.
diff --git a/ElectronJS/config.json b/ElectronJS/config.json
@@ -1,7 +1 @@
-{
- "webserver_address": "http://localhost",
- "webserver_port": 8074,
- "user_data_folder": "./user_data",
- "debug": true,
- "absolute_user_data_folder": "D:\\Document\\Projects\\EasySpider\\ElectronJS\\user_data"
-}
+{"webserver_address":"http://localhost","webserver_port":8074,"user_data_folder":"./user_data","debug":true,"absolute_user_data_folder":"D:\\Document\\Projects\\EasySpider\\ElectronJS\\user_data"}
diff --git a/ElectronJS/main.js b/ElectronJS/main.js
@@ -457,7 +457,7 @@ function handleOpenBrowser(event, lang = "en", user_data_folder = "", mobile = f
  runBrowser(lang, user_data_folder, mobile);
  let size = screen.getPrimaryDisplay().workAreaSize;
  let width = parseInt(size.width);
- let height = parseInt(size.height * 0.65);
+ let height = parseInt(size.height * 0.6);
  flowchart_window = new BrowserWindow({
  x: 0,
  y: 0,

diff --git a/ElectronJS/src/taskGrid/FlowChart_CN.html b/ElectronJS/src/taskGrid/FlowChart_CN.html
@@ -513,25 +513,33 @@ <h4 class="modal-title">等价XPath</h4>
  <h4 class="modal-title" id="myModalLabel">保存任务</h4>
  <button type="button" class="close" data-dismiss="modal" aria-hidden="true">&times;</button>
  </div>
- <div class="modal-body">
+ <div class="modal-body" style="height:400px;overflow: auto">
  <input onkeydown="inputDelete(event)" id="serviceId" type="hidden" name="serviceId" value="-1"></input>
  <input onkeydown="inputDelete(event)" id="url" type="hidden" name="url" value="about:blank"></input>
  <label>任务名称：</label>
  <input onkeydown="inputDelete(event)" required name="serviceName" value="新web采集任务" id="serviceName" class="form-control"></input>
  <label>任务描述：</label>
  <input onkeydown="inputDelete(event)" id="serviceDescription" name="serviceDescription" class="form-control"></input>
- <label>每采集多少条数据保存一次（值越大采集速度越快，但如果意外退出则有数据丢失风险）：</label>
- <input onkeydown="inputDelete(event)" type="number" value="10" id="saveThreshold" name="saveThreshold" class="form-control"></input>
- <label>是否为cloudflare等极端反爬网站：</label>
- <select id="cloudflare" name="cloudflare" class="form-control">
- <option value = 0>否</option>
- <option value = 1>是</option>
+ <label>导出数据格式：</label>
+ <select id="outputFormat" class="form-control">
+ <option value = "csv">CSV</option>
+ <option value = "xlsx">XLSX（EXCEL）</option>
+ <option value = "mysql">MySQL</option>
  </select>
  <label>浏览器模拟类型：</label>
  <select id="environment" name="environment" class="form-control">
  <option value = 0>电脑端</option>
  <option value = 1>手机端（Cloudflare模式下不支持）</option>
  </select>
+ <label>是否为cloudflare等极端反爬网站：</label>
+ <select id="cloudflare" name="cloudflare" class="form-control">
+ <option value = 0>否</option>
+ <option value = 1>是</option>
+ </select>
+ <label>每采集多少条数据保存一次（值越大采集速度越快，但如果意外退出则有数据丢失风险）：</label>
+ <input onkeydown="inputDelete(event)" type="number" value="10" id="saveThreshold" name="saveThreshold" class="form-control"></input>
+ <label>控制台预览时数据最大显示长度：</label>
+ <input onkeydown="inputDelete(event)" type="number" value="15" id="maxViewLength" class="form-control"></input>
 
  </div>
  <div class="modal-footer">

diff --git a/ElectronJS/src/taskGrid/logic_CN.js b/ElectronJS/src/taskGrid/logic_CN.js
@@ -427,6 +427,8 @@ function saveService(type) {
  "saveThreshold": saveThreshold,
  "cloudflare": cloudflare,
  "environment": environment,
+ "maxViewLength": parseInt($("#maxViewLength").val()),
+ "outputFormat": $("#outputFormat").val(),
  "containJudge": containJudge,
  "desc": serviceDescription,
  "inputParameters": inputParameters,
@@ -460,10 +462,24 @@ if (sId != null && sId != -1) //加载任务
  $.get(backEndAddressServiceWrapper + "/queryTask?id=" + sId, function(result) {
  nodeList = result["graph"];
  app.$data.list.nl = nodeList;
+ for(let node of nodeList){ //兼容旧版本
+ if(node["option"] == 1){
+ if(!("cookies" in node["parameters"])) {
+ node["parameters"]["cookies"] = "";
+ }
+ }
+ }
  $("#serviceName").val(result["name"]);
  $("#serviceId").val(result["id"]);
  $("#url").val(result["url"]);
  $("#serviceDescription").val(result["desc"]);
+ for(let key of Object.keys(result)){
+ try{
+ $("#"+key).val(result[key]);
+ } catch(e){
+ console.log(e);
+ }
+ }
  refresh();
  });
 } else {

diff --git a/ElectronJS/tasks/1.json b/ElectronJS/tasks/1.json
diff --git a/ElectronJS/tasks/112.json b/ElectronJS/tasks/112.json
diff --git a/ElectronJS/tasks/142.json b/ElectronJS/tasks/142.json
@@ -1 +1 @@
-{"id":142,"name":"京东全球版-专业的综合网上购物商城","url":"https://www.jd.com","links":"https://www.jd.com","create_time":"7/6/2023, 3:38:35 AM","version":"0.3.5","saveThreshold":10,"cloudflare":0,"environment":0,"containJudge":false,"desc":"https://www.jd.com","inputParameters":[{"id":0,"name":"urlList_0","nodeId":1,"nodeName":"打开网页","value":"https://www.jd.com","desc":"要采集的网址列表，多行以\\n分开","type":"string","exampleValue":"https://www.jd.com"}],"outputParameters":[{"id":0,"name":"参数1_文本","desc":"","type":"string","exampleValue":"/手机/数码"}],"graph":[{"index":0,"id":0,"parentId":0,"type":-1,"option":0,"title":"root","sequence":[1,2],"parameters":{"history":1,"tabIndex":0,"useLoop":false,"xpath":"","wait":0},"isInLoop":false},{"id":1,"index":1,"parentId":0,"type":0,"option":1,"title":"打开网页","sequence":[],"isInLoop":false,"position":0,"parameters":{"useLoop":false,"xpath":"","wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"url":"https://www.jd.com","links":"https://www.jd.com","maxWaitTime":10,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"cookies":"test=123\nipLoc-djd=53283-53456-0-0\nareaId=53283\nmba_sid=16885856346417163685425076773.0\n__jdc=122270672\n__jdb=122270672.1.16885856346381587112207|1.1688585634\nmba_muid=16885856346381587112207\n__jdv=122270672%7Clocalhost%3A8074%7C-%7Creferral%7C-%7C1688585634639\n__jda=122270672.16885856346381587112207.1688585634.1688585634.1688585634.1"}},{"id":2,"index":2,"parentId":0,"type":1,"option":8,"title":"循环","sequence":[3],"isInLoop":false,"position":1,"parameters":{"history":4,"tabIndex":-1,"useLoop":false,"xpath":"/html/body/div[5]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":1,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":["/html/body/div[5]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]","//div[contains(., '/手机/数码')]","//DIV[@class='LeftSide_menu_item__SBMWC LeftSide_text_space__2UhbG ']","/html/body/div[last()-5]/div/div[last()-4]/div/div[last()-2]/div/div/div/div[last()-1]/div[last()-12]"]}},{"id":3,"index":3,"parentId":2,"type":0,"option":3,"title":"提取数据","sequence":[],"isInLoop":true,"position":0,"parameters":{"history":4,"tabIndex":-1,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"paras":[{"nodeType":0,"contentType":0,"relative":true,"name":"参数1_文本","desc":"","extractType":0,"relativeXPath":"","allXPaths":"","exampleValues":[{"num":0,"value":"/手机/数码"}],"unique_index":"p2h2i1dva8ljq4aje2","iframe":false,"default":"","beforeJS":"","beforeJSWaitTime":0,"JS":"","JSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"downloadPic":0}],"loopType":1}}]}
+{"id":142,"name":"京东全球版-专业的综合网上购物商城","url":"https://www.jd.com","links":"https://www.jd.com","create_time":"7/6/2023, 4:08:31 AM","version":"0.3.5","saveThreshold":10,"cloudflare":0,"environment":0,"maxViewLength":15,"outputFormat":"xlsx","containJudge":false,"desc":"https://www.jd.com","inputParameters":[{"id":0,"name":"urlList_0","nodeId":1,"nodeName":"打开网页","value":"https://www.jd.com","desc":"要采集的网址列表，多行以\\n分开","type":"string","exampleValue":"https://www.jd.com"}],"outputParameters":[{"id":0,"name":"参数1_文本","desc":"","type":"string","exampleValue":"/手机/数码"}],"graph":[{"index":0,"id":0,"parentId":0,"type":-1,"option":0,"title":"root","sequence":[1,2],"parameters":{"history":1,"tabIndex":0,"useLoop":false,"xpath":"","wait":0},"isInLoop":false},{"id":1,"index":1,"parentId":0,"type":0,"option":1,"title":"打开网页","sequence":[],"isInLoop":false,"position":0,"parameters":{"useLoop":false,"xpath":"","wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"url":"https://www.jd.com","links":"https://www.jd.com","maxWaitTime":10,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"cookies":"test=123\nipLoc-djd=53283-53456-0-0\nareaId=53283\nmba_sid=16885856346417163685425076773.0\n__jdc=122270672\n__jdb=122270672.1.16885856346381587112207|1.1688585634\nmba_muid=16885856346381587112207\n__jdv=122270672%7Clocalhost%3A8074%7C-%7Creferral%7C-%7C1688585634639\n__jda=122270672.16885856346381587112207.1688585634.1688585634.1688585634.1"}},{"id":2,"index":2,"parentId":0,"type":1,"option":8,"title":"循环","sequence":[3],"isInLoop":false,"position":1,"parameters":{"history":4,"tabIndex":-1,"useLoop":false,"xpath":"/html/body/div[5]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"loopType":1,"pathList":"","textList":"","code":"","waitTime":0,"exitCount":0,"historyWait":2,"breakMode":0,"breakCode":"","breakCodeWaitTime":0,"allXPaths":["/html/body/div[5]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]","//div[contains(., '/手机/数码')]","//DIV[@class='LeftSide_menu_item__SBMWC LeftSide_text_space__2UhbG ']","/html/body/div[last()-5]/div/div[last()-4]/div/div[last()-2]/div/div/div/div[last()-1]/div[last()-12]"]}},{"id":3,"index":3,"parentId":2,"type":0,"option":3,"title":"提取数据","sequence":[],"isInLoop":true,"position":0,"parameters":{"history":4,"tabIndex":-1,"useLoop":false,"xpath":"","iframe":false,"wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"paras":[{"nodeType":0,"contentType":0,"relative":true,"name":"参数1_文本","desc":"","extractType":0,"relativeXPath":"","allXPaths":"","exampleValues":[{"num":0,"value":"/手机/数码"}],"unique_index":"p2h2i1dva8ljq4aje2","iframe":false,"default":"","beforeJS":"","beforeJSWaitTime":0,"JS":"","JSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"downloadPic":0}],"loopType":1}}]}
diff --git a/ElectronJS/tasks/143.json b/ElectronJS/tasks/143.json
@@ -0,0 +1 @@
+{"id":143,"name":"中国知网","url":"https://chn.oversea.cnki.net/index/","links":"https://chn.oversea.cnki.net/index/","create_time":"7/6/2023, 4:50:52 AM","version":"0.3.5","saveThreshold":10,"cloudflare":0,"environment":0,"maxViewLength":15,"outputFormat":"csv","containJudge":false,"desc":"https://chn.oversea.cnki.net/index/","inputParameters":[{"id":0,"name":"urlList_0","nodeId":1,"nodeName":"打开网页","value":"https://chn.oversea.cnki.net/index/","desc":"要采集的网址列表，多行以\\n分开","type":"string","exampleValue":"https://chn.oversea.cnki.net/index/"}],"outputParameters":[],"graph":[{"index":0,"id":0,"parentId":0,"type":-1,"option":0,"title":"root","sequence":[1],"parameters":{"history":1,"tabIndex":0,"useLoop":false,"xpath":"","wait":0},"isInLoop":false},{"id":1,"index":1,"parentId":0,"type":0,"option":1,"title":"打开网页","sequence":[],"isInLoop":false,"position":0,"parameters":{"useLoop":false,"xpath":"","wait":0,"waitType":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"url":"https://chn.oversea.cnki.net/index/","links":"https://chn.oversea.cnki.net/index/","maxWaitTime":10,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"cookies":""}}]}